The present invention relates to a process for the production by catalysis of a layer having a high planar magnetic anisotropy in a ferrimagnetic garnet. It more particularly applies to the field of producing magnetic bubble stores and particularly non-implanted disk bubble stores, as well as in the field of producing magneto optical or semiconductor material.
In general terms, the production of a bubble store firstly consists of producing by epitaxy a ferrimagnetic garnet layer with growth anisotropy perpendicular to the layer on an amagnetic substrate, mainly a garnet. It is pointed out that magnetic bubbles are small magnetic domains, whose magnetization, directed perpendicular to the surface, is reversed compared with that of the material containing the bubbles. The ions are then implanted in the epitactic layer.
This ion implantation makes it possible to produce on the surface of the ferrimagnetic garnet layer a planar magnetization layer, i.e. a layer whose magnetization is parallel to the surface of said layer. This planar magnetization layer has the object of increasing the stability of the magnetic bubbles. This ion implantation makes it possible to produce planar magnetization layers over a thickness of approximately 0.5 .mu.m.
By using an appropriate implantation mask, it is possible to define in the case of bubble stores with non-implanted patterns, propagation patterns, which are contiguous patterns, having the shape of a disk, lozenge, etc. As ion implantation is only carried out around these patterns, the latter are called non-implanted patterns.
In the case of bubble stores with patterns based on iron and nickel, ion implantation, apart from serving to form the surface layer with planar magnetization, is also used for eliminating the "hard" bubbles, i.e. the bubbles having structures with complex walls.
The propagation of the magnetic bubbles along the propagation patterns is realised by applying a rotary d.c. field in a direction parallel to the surface of the ferrimagnetic layer. The bubbles positioned below the planar magnetization surface layer are bonded to non-implanted propagation patterns via a potential well due to the stress field between the implanted and non-implanted zones. The displacement of the magnetic bubbles along the propagation patterns results from the action of the rotary field, which produces a mobile charged wall entraining the bubbles.
For a considerable time use has been made of the magnetostriction properties of the ferrimagnetic garnet layers to obtain said magnetic anisotropy of the surface layer. Thus, ion bombardment produces on the surface of the epitactic garnet layer, defects which consequently lead to a deformation of the mesh parameter in the direction perpendicular to said ferrimagnetic garnet layer. Within the garnet layer, said defects produce high mechanical stresses oriented parallel to the surface of said layer. It has been proved that an expansion of the mesh parameter could not be carried out parallel to the surface of the ferrimagnetic layer.
The ferrimagnetic garnet layers are produced so as to have a negative magnetostriction coefficient. In this case, a compressive stress obtained by ion implantation induces magnetic anisotropy in the plane of the implanted surface layer which exceeds the growth anisotropy of the starting material, i.e. the non-implanted material.
Unfortunately this magnetostriction mechanism has limits depending on the size of the growth anisotropy of the material (growth by epitaxy), as well as its negative magnetostriction coefficient. Thus, it is not possible to increase the implanted ion dose indefinitely, because beyond a certain threshold of the defects, the magnetism of the implanted surface layer is cancelled out and it is no longer possible to move the bubbles along the non-implanted propagation patterns.
However, in view of the fact that new generations of magnetic bubble stores and in particular non-implanted pattern stores tend to store ever higher information densities, it is necessary for ever decreasing sizes of the magnetic bubbles, which cannot be achieved using a material with a high growth anisotropy. Unfortunately, with such materials, it is no longer possible to obtain a planar magnetization in the implanted layer by a simple magnetostriction mechanism.
In order to increase the magnetic anisotropy of the implanted layer, no matter what the growth anisotropy of the starting material, consideration has recently been given to carrying out a reverse sputtering of argon ions in said implanted layer. This is carried out by heating a sample to above 100.degree. C. This process is described in the article entitled "Magnetic and Crystalline Properties of Ion-implanted Garnet Fibres with Plasma Exposure" by K. Betsui et al, published at the Intermag Conference, Hamburg in 1984.